# Acquire the source

## Task and place

Acquiring the source is the first hands-on stage of a formalization. A named act becomes preserved bytes plus a provenance record, and every later stage stands on that record: pinning re-verifies these bytes, inventory reads them article by article, and rules cite extracts taken from them. If the bytes are wrong here, every hash downstream certifies the wrong text, convincingly.

## Inputs

- The identity of the act: title, number, adopting body, and the date of the wanted reading.
- Candidate publications: publisher feeds, official gazettes, office copies, and any structured feed the publisher offers.
- The [scope card template](/handbook/files/templates/scope-card.md), which will carry the scope boundary and the provenance note.
- A place inside the package for original files, kept apart from anything written by hand.

## Actions

- Choose one publication as the base and write down why it won over the alternatives.
- Record the provenance status honestly: a publisher original, an office copy, or an unofficial copy are different claims, and the record must carry the true one.
- Record the retrieval moment and the address the bytes came from, exactly as used.
- Preserve the original bytes unchanged next to the package. Never hand-edit them, not even to fix an obvious typo.
- Note the media of what arrived: plain text, a paged document, a scanned image with recognized text, or a structured feed. Each conversion gets a journal line naming the tool, its settings, and what was checked afterwards.
- Collect attachments and annexes that carry normative content. Tariff tables and forms often live outside the main text.

## Decisions

- Which publication is the base, and which honesty label it carries. An unofficial copy stays labeled as one; the label used below is literal.
- Where the scope boundary runs: which articles enter this package version and which stay out.
- Which language version is the base when the publisher offers several, and which others are kept alongside.

## Artifact

The artifact of this stage is a source card plus a transform journal: one card per base publication, each conversion step journaled or explicitly marked as absent. Start from the [scope card template](/handbook/files/templates/scope-card.md); the filled [EAI scope card](/handbook/files/filled/eai-scope-card.md) shows the shape for the running example.

## Worked example

The running example through this handbook is a small insurance package, versioned and dependency-free with explicit local imports, covering five articles of an employer accident-insurance act: the employer duty to insure, the insurer penalty, the assumed insured-sum input, the tariff table with the premium formula, and the payout condition.

Its base publication is the Russian text of the act obtained from the publisher document feed and retrieved in mid-September. The provenance status is recorded literally as `PINNED_UNOFFICIAL_COPY`: a pinned copy, honestly not the official publication. The original bytes and the feed envelope sit next to the package; the envelope names the publisher document id and a feed version date in late August. The observed file listing:

```text
sources/employee-accident-insurance:
adilet-api.json
ru.txt
```

## Pitfall

The common failure is polishing the text before hashing: fixing typos, re-wrapping lines, normalizing quotes. Every later check then certifies polished bytes while claiming the act. A quieter variant is re-downloading from the feed when the publisher updates it, which swaps the bytes under an already recorded hash. Preserve first, hash second, and treat any later byte change as a new acquisition with its own journal line.

## Verification

The stage is done when each of the following holds:

- The source card names the publication, the provenance status, the retrieval moment, and the original address.
- The original bytes are present next to the package, byte-identical to what was retrieved.
- Every conversion has a journal line, or the card states that there was none.
- The scope boundary lists included and excluded articles by number, with no article silently missing.

## Limits

Acquisition says nothing about legal force and nothing about which reading applies at a given date. It also does not verify that formalized extracts occur in the bytes; that bond is drawn and re-checked at pinning.

## Next step

Continue with [Pin the edition](/handbook/pin-editions/), which binds the formalization to the bytes acquired here.

## Sources

- [Source and anchor](/tutorials/real-article-source/) — why an address in the act is not yet a link to its bytes.
- [Reconciliation with the pinned edition](/tutorials/pinned-edition/) — what the presented document must look like at check time.